This study was initiated to provide information about chemical structures as an aid in selecting from potential acquisitions, compounds for screening. Statistics on structure-activity relations over broad classes of structures are being used to predict the probability of antitumor activity of new compounds in the mouse lymphocytic leukemia (P388) screen. Also, a measure of rarity or uniqueness of a new structure is derived from the statistics on the structural characteristics of compounds that have been adequately screened in P388, regardless of their activity. It is planned to integrate the P388 model into DTP Chemical Information System for use operationally to aid in the selection of compounds for screening. BIBLIOGRAPHIC REFERENCE: Hodes, L., Hazard, G. F., Gerna, R., Richman, S.: A Statistical-Heuristic Method for Automated Selection of Drugs for Screening. J. Med. Chem., 20: 469-475, 1977.